Jeremy's dashboard

In the past we've put aside the "I hate it" concept for saying particular movies are so bad you never want to rank them again, because it sidesteps the one-on-one ranking mechanism of the site. In theory, the ranking system should be able to tell you your least favorite movies as well as your favorite ones... but since no one spends much time at the bottom of their list, it's not as effective in practice. We're working on ways to address that.

I actually think the notion of a "Movie Basement" is pretty cool. In the near future we're going to let you create custom lists of your own that as you rank you can add movies to. We'll probably start with a few built-in ones, and the Movie Basement concept could be one of them. I agree that it's pretty cool to be able to see not just the movies a user loves, but those s/he deems to be among the worst dregs of all time.

Thanks for the suggestion!

Hey guys -

Big apology for the slowness lately. There were a few bugs in that last release we pushed, and they were erratic and hard to track down. There was one logical error and a problem with some of the movie data, and combined they just killed the experience for some users.

Mind giving it another shot? I took a look at the results that should be coming back now for ghosthand27, ottobud, and some others, and it was all returning good movie pairs. Let us know here if you continue to see this particular issue.

John,

Definitely. The iPhone site has unfortunately suffered slow development lately as we've tried to keep up with performance and features on the desktop site, but we hope to return to it in the near future to at least implement some core things like the filters and a few other things.

I wish the phone site already had filters, friends, discussions, movie stats, trailers, and more -- just tough to find the time! Thanks for hanging in there. We'll do our best to round it out soon.

John,

We don't have a way for you to manually manipulate the entire list as such, but if you go to your favorite movie's info page (either do a search at the top of the site or click on the title in your list), you can click Re-Rank This Movie to get several immediate matchups for it. It'll rise the ranks quickly. (Note that you can reload the page to do it again when you're done.)

FYI, you can do the same thing for any movie on the charts page (http://www.flickchart.com/charts) by clicking Rank This Movie there, too.

Thanks a lot for the compliment -- we really appreciate it.

Darksidebitca - Haha, yeah. We thought it would be a curious little tidbit for everyone. And yeah, it does of course assume a single time through -- originally I played around with explicitly saying that, but I decided it's obvious enough as it is.

We've toyed with notions of indicating when you saw a movie, and I'm wondering about also letting users say how many times they've watched it (e.g. "Once", "Twice", "A Handful of Times", "Lots", or "ZOMG you have no freaking idea"). Not sure how many users would be into that, but it would let us do some cool stuff (including better ranged estimates of your time watched -- though that would just scratch the surface).

Joseph (and Alex),

There's definitely a shortage of movies, that's for sure. We're trying to balance upgrading the servers & streamlining the traffic to keep the site running smoothly (there have been times where it definitely hasn't), along with new features, the completely rewritten algorithm, and fixing and augmenting the movie data.

Right now the movie data and the algorithm are at the top of the list. Sorry for the wait -- we know that a lot of people's favorite movies are on the hit list.

They weren't omitted because of priority or anything like that -- there were unfortunately just sporadic errors and missing data when we meshed all of it together from different sources in the first place. Our big task now is to fill in the gaps for the movies we have in the database (most of the big movies you'd mention), and add onto it from new sources (2008/2009, for instance, are also way too light).

Please bear with us (and I know, we say this about so many things!) -- we're getting there.

Kopiok,

I do know of at least a couple like this (Il postino, the Italian movie, comes to mind first). Sometimes the foreign title is carried over to the US/UK/CA/etc., which is why it's kept. In the case of Il postino, the title was extended to be "Il Postino: The Postman", so I could argue that we should use that title, although it's a bit unwieldy. I'd probably take them on a case-by-case basis.

Do you remember any in particular?

Adam,

Good ideas all. We've actually got a handful of threads here on GS about our plans for friends, grouping, stats, recommendations, and comparisons -- but it's not always that easy to find with the search tool here. We've got big plans for that, though.

We also have plans for filtering on your favorite movies, the films you haven't seen, and the global charts. And in the much nearer term, too -- we were going over our half-done page layout just last night for that. Hold on, it's coming!

Problem is, I'm going to be ranking more and filtering my own lists instead of coding the next feature for at least a couple hours. But I could waste way longer on the friends thing, so I'd say the ordering is ideal for continued progress.

Jaydon & Ted - Yeah, it's our site, but we want you guys to have fun with it. We could've just made it for ourselves and locked it away otherwise.

Absolutely agreed on gauntlet and easier/better re-ranking and pairing of movies. All that's coming along with the new algorithm, and there's high probability of it kicking butt.

Otto,

That's an interesting pattern about the remakes. I've seen it here and there myself, but I didn't realize that particular motif was so widespread. A lot of people have been confused about why particular movies are missing, and with good reason, since it's not communicated very well in a central place.

The movies we have generally aren't there through intentional discrimination; we started with a large set of data, did some manipulation to it, and tried to import it all; but the data comes from a lot of places and gets meshed together, and there are a lot of movies that had incomplete or incorrect data as a result.

We're trying to sort all that out now so we can correct our original database and supplement it with lots of additional films. Sorry for the confusion and inconvenience in the meantime, but thanks for bearing with us.

Ted -

No, you're right. Specific matchups are more for sharing and discussing than to correct your list.

Being able to re-rank an individual movie, as we currently allow from the movie info page, helps with your second-biggest problem, but it's meant more as a way to:

a) launch an unranked movie into your chart at a good spot
b) to let you have immediate fun with a favorite movie that hasn't come up much yet

... again, not primarily as a list-fixer.

To the extent that you can use either of those to "solve" problems in your list, they're covering up the real problem, which is the way the current algorithm works. As you say, if you get a bad movie up really high, it can take a long time for it to "bubble" down as you rank better movies above it. The higher up it is to start with, the disproportionately longer it takes, since it has lower odds of being beaten directly or indirectly in a matchup.

The new algorithm keeps much better track of past rankings, and builds all your matchups into a much more complete body of knowledge, so that big shifts in the list will take a single click, rather than doing it a single movie at a time. I think you'll like it.

I'm pretty confident the new system will sublimate your need for shuffling stuff around. We won't implement that, because it bypasses the entire system of ranking that the site's set up to enable. It's meant not as just another equivalent way to move titles around a list, but to change the game a bit by:

a) making it a lot more fun than sitting down and poring over a list of names
b) stumbling upon movies you wouldn't have remembered to look up
c) finding movies you've never seen before that look interesting
d) making those hard choices at the top more personal and in focus

Enabling drag n' drop would skip past all of those goals in an instant and become like every other listmaking site out there. Maybe you hate the clicking, maybe you hate the matchups, in which case it's true, maybe Flickchart isn't for you.

But I hope our forthcoming changes will make all the difference. Stay tuned.

Actually, this is a subtle but very good point, and it gets into the finer aspects of movie categorization along the lines of the music genome (although not that complex).

Most movies belong to several genres, with the first one or two being primary. The ones after that, if they're not subgenres or relatives of those first couple (like horror and thriller), probably won't accurately characterize the movie.

Take Clue. To look at an independent source, IMDB lists it as Comedy-Crime-Mystery-Thriller. All elements, to be sure. But let's look at thriller - what movies do we really think of for that genre?

Some of the top ones are: The Godfather, Se7en, Rear Window, Alien, The Silence of the Lambs, North by Northwest, Memento, and so on. Not too many comedies in there. I might think Clue is a really endearing movie that's top 20 for me, but when I'm in the mood for thrillers, I'm not reaching for Clue.

The genre - your context - does a lot to set the tone and the expectation.

This also has importance for recommendations, because Clue might be highly rated overall, but we give you the wrong result if we follow the logic:

- You love thrillers (the top of your list is full of them)
- Clue is a thriller
- Clue is really highly rated
- You haven't seen Clue
- You'll love Clue

Therefore, it would be sensible to either do as you say and allow ranking within genres to not impact the general rankings; or (and this is the option I prefer, to get the best of both worlds) allow a "strength of association" for movie tags as we implement them, based on the community. In other words, each tag would have a weight relative to a movie: Clue would be strongly bound with Comedy, but weakly with Thriller.

That way we can go either way with it later: make the genre filters strict or loose, or allow the user some measure of control over it (if they want to discover more movies in a particular genre, even if they're not as hardcore set in that genre).

Make sense?

Jaydon - It'll be *pretty* smart about it. :)

You'll still have to endure some matchups with it - especially if you were to start over and get TDK in place right near the beginning - but it'll be way fewer than the current kludgy algorithm dishes out.

When it does come up, it will at least be against much more worthy movies. Not A Walk in the Clouds starring Keanu Reeves.

Derek - Sorry to hear it. What exactly are you seeing when you get to the site? Do you see the spot for username/password/email, but it won't let you create the account? Do you get a particular error?

Adam - Haha, glad you're enjoying it.

I know what you're saying about the algorithm, since I thought the same thing plenty of times when I was originally thinking about it. Because the system is constantly refining your list, the two cases (20-21 vs. 100-101) are essentially equivalent to it; it doesn't know for a fact which of the two is more out of place. (It technically gains knowledge the more you rank so that it can make better statistical guesses, but the current algorithm is somewhat poor at that.) So you're right -- either case can be preferable, depending on circumstances.

So the question then is how to decide which to do. It seems like a lot of technical complexity to burden the user with -- even if s/he decides not to do it each time, it's one of those things that might weigh on your mind (am I doing it right? should I move those down this time? it's close. wait, lemme check what else is down at 100). And we definitely don't want to drag down the ease of the experience - we want it to primarily be fun, with a good list being the by-product without having to think heavily about it.

Granted, it can be said that a good list is not *currently* the by-product anyway (or that it takes way too long), but we're working on it. :P

I think the new algorithm will work sufficiently quickly and well - with a lot more movement than you're seeing now - that you won't feel the need for manual intervention. The system should hopefully be taking care of it for you.

I'd love to hear back from you once it's up (hopefully within a week or so).

Haha, hopefully it was fixed as of this morning (although like I said, you would definitely have seen it once more, so it could set the cookie). Is that not the case? If it's not, sorry. I guess in that scenario there are hundreds of people sick of that matchup by now.

Sinistar - More than thoughts. And SURELY there's even more out there than just movies and music, wouldn't you say? :P

Tony - Awesome, really glad to hear that it made a difference. We made a change to the database a few days ago. It's been a lot faster for us, but you're the first feedback to say that it worked for people at large. Thanks for the confirmation!